Skip to content

Exoskeletons Won’t Happen

Lump them in with flying cars, jet packs, and nuclear-powered toothbrushes because exoskeletons just aren’t a good enough idea to be worth building. Particularly the military exoskeleton.


If the exoskeleton can move a person faster than they could normally move, that means the suit is moving the person. Since arms and legs reverse direction constantly, that means the suit is “shaking the baby” or beating on the person and they’ll be battered into a soggy pulp. If the exoskeleton can give a person access to more force then they would normally have, at the same speeds, then it’s an extremely complicated servo that has to carry the operator along while shielding them from the forces involved.

So we’re not going to get the first option working. As soon as the exoskeleton moves faster than the person it can no longer use the person’s movements as reliable control inputs. It’s like how a cat tries to follow you but walks in front of you (cats are stupid). There’s a chance we might be able to link the exoskeleton directly to the brain, but then the person’s limbs become delicate dead weight and might as well be tucked into the fetal position (hands and arms inside the vehicle).

We have made the second option work already, more or less. However, outputting force through a system that has to be wrapped around, and not interfere with, a human body severely compromises the power train. Even the defense industrial establishment can’t make an exoskeleton that reduces operator effort (pdf), but there’s a steadily increasing list of people building what amounts to a weak crane and strapping it to their body. I’ve even seen a couple different bicycle-powered forklifts. So the more like an actual vehicle the exoskeleton is the better it performs per dollar invested. Again, might as well just tuck the person into a little hole in the middle and keep all their limbs in one place.

Additionally, assuming we do make progress on the phenomenal complexity necessary for an exoskeleton to sort of work by performing so well that it overcomes those limitations, at that point the important question becomes why don’t we just build an android? Using the same parts, an android would be a lot higher value than an exoskeleton, assuming remote control wasn’t an issue (in most cases it wouldn’t be). If we can build an android, then why not just use the same parts and dollars to build a couple simpler robots?

There just isn’t much overlap between 1) needing to be shaped exactly like a human, 2) needing to be absurdly strong, and 3) having no control over the situation. If I remember correctly, this scene from Pumping Iron was captured because they were wandering around trying to find something heavy for Franco Columbu to lift in the real world and the only thing they found was a car that was awkwardly parked.

Out in the real world vast strength isn’t particularly useful, especially when it’s shaped like a person. A person can lift a car straight up, but if they want to move it they have to walk, which is a super bad idea for something shaped like a human when holding a heavy thing out in front. A much better idea would be to put a jack on a dolly, lift the car, and push the dolly. In the real world that’s totally an option and it’s superior to having a general purpose strongman wandering around. In the real world we can break a big job into smaller pieces. Here’s some random guy moving buildings and Stonehenge-sized rocks all by himself.

So, an exoskeleton could be useful if a person is all alone and has to do something requiring phenomenal strength in a short period of time and for whatever reason they have to use one tool shaped like a human. Remove any of these constraints and there’s no reason to even consider an exoskeleton.

So the general idea just isn’t very good and the specific military application doesn’t seem to add anything. Soldiers always work in teams and they have access to a lot more options for solving problems than the suit that one guy is wearing. The broader goal is that special operators want to be bullet proof, which means lots of armor, and the ordinary operators want to be well supplied, which means lots of kitchen sinks. So the real underlying goal is moving heaving things around like they’re not heavy, the smaller and more precise the movements the better.

A whole industry is already doing this: the construction industry. Construction equipment is surprisingly precise. You can find a lot of cool videos of things like excavators picking up coins.

The thing about construction equipment is that it’s specialized (because there’s no reason to pretend that the tools has to look like a human and all work has to be done in one single action). It’s not just that the equipment itself is heavy, it’s that if you want it to exert controlled force at a distance the mechanism in between has to be simple. So if you have different jobs you need different simple mechanisms for them. There’s no point using something as complicated as a human limb to directly control a simple mechanism, so their controls are simple. Once you decouple the human body from the control you can get away with amplifying the inputs, because the body itself doesn’t have to follow along with the result.

So, what the military should be developing are a lot of weird little one-person vehicles for specific activities. Even if they do come up with an exoskeleton that works adequately the individual components will immediately be applied to simpler vehicles with greater success. The inherent system-level compromises of an exoskeleton will just suck to much of the theoretical performance out of the subsystems.

Purely from a perspective of military strategy an exoskeleton isn’t going to be a good value. The most obvious need for strength is carrying armor, but the armor severely limits range of motion, negating the idea of shaping the vehicle like a human. One could carry heavier weapons or more gear, but then they become an attractive target that’s easy to hit because it’s big and doesn’t have enough armor to protect itself. An exoskeleton will never be cool, quite or fully shielded in our lifetimes, meaning that the emissions it gives off (IR, sound, EM, etc) will negate any attempt to use it in special operations roles. If it gets even a little bigger than an average human it will start to get tangled up in doorways, stairs and furniture. If it gets too heavy it will start to slip on and fall through the surfaces that are too weak to support it.

I think the closest we might get to an exoskeleton with legs would look kind of like a clown, with big feet and a fat belly. The feet would be wheels or tracks that would do 99% of moving the thing around. The legs would mostly be used as shock absorbers and only used to reposition the wheel-feet when necessary on complicated terrain. The belly would be a half-cockpit where the operator would keep their hands on a control panel. The top would probably be a swappable “turret” that would be changed out for different missions. But even that is probably not useful in enough situations.

The closest I think we will actually get to an exoskeleton is a vehicle with arms. Legs are a bad foundation for getting work done. Ideally you’d just put the thing somewhere stable and leave it. If it needs to be able to move itself then the sturdiest option is tracks or wheels, no suspension. Those two options can deal with a surprising amount of terrain. If the vehicle does need to pass through something where tracks/wheels don’t work it can just use its arms to move itself, or the obstacles, and then keep rolling.

There’s a small chance, way in the future, that exoskeletons might add value in a few niche situations. For example, a suit of powered armor could be useful for a particularly well-funded personal security detail. Nothing like letting people know up front that you’re invulnerable in a flashy way to get into the right clubs. In that case the impracticality would be part of the appeal.

Published infuture techsystems engineering

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *